🗳️🇨🇴 Invamer just published Colombia Opina #21. Read the headlines and you'd think Cepeda has this election locked 🐾
he leads first round at 44%, he wins every simulated second round, and the trendline since 2024 favors him. That's the story Caracol and BluRadio are selling tonight.
The actual data inside the 60-page report tells a very different story. Once you stop reading the cross-tabs the way the pollster wants you to read them, the picture flips.
he leads first round at 44%, he wins every simulated second round, and the trendline since 2024 favors him. That's the story Caracol and BluRadio are selling tonight.
The actual data inside the 60-page report tells a very different story. Once you stop reading the cross-tabs the way the pollster wants you to read them, the picture flips.
Time to talk methodology, because this is where Invamer's structural bias becomes visible.
The technical sheet reports a non-response rate of 72.50%!!!! LOL 😂
For every four doors the surveyors knocked, three people refused to participate. Treat that number as the single most important piece of metadata in the entire document 💩. Everything else flows from it.
At a 72.5% refusal rate, the 3,800 published respondents do not constitute a random cross-section of Colombian voters, they are a self-selected pool of people willing to invite a stranger into their home and answer political questions in person, and that pool has predictable demographic and political tilts
The voters most likely to decline an in-home interview of that length are working-age professionals with time constraints, men in their 30s and 40s, urban dwellers with privacy concerns, and voters who distrust the mainstream media outlets funding the survey. Those are exactly the segments where Paloma and Abelardo over-perform.
ISO 20252 certification doesn't fix this problem. It just documents that the bias is being introduced in compliance with the standard.
The technical sheet reports a non-response rate of 72.50%!!!! LOL 😂
For every four doors the surveyors knocked, three people refused to participate. Treat that number as the single most important piece of metadata in the entire document 💩. Everything else flows from it.
At a 72.5% refusal rate, the 3,800 published respondents do not constitute a random cross-section of Colombian voters, they are a self-selected pool of people willing to invite a stranger into their home and answer political questions in person, and that pool has predictable demographic and political tilts
The voters most likely to decline an in-home interview of that length are working-age professionals with time constraints, men in their 30s and 40s, urban dwellers with privacy concerns, and voters who distrust the mainstream media outlets funding the survey. Those are exactly the segments where Paloma and Abelardo over-perform.
ISO 20252 certification doesn't fix this problem. It just documents that the bias is being introduced in compliance with the standard.
Look at who actually responded. From the demographics section:
- Age 55 and over: 26.3% of the sample
- Strata 1 and 2 combined: 51.4%
- Homemakers, retirees, and unemployed combined: 30.0%
- Rural: 22.4%
This sample skews older, lower-strata, and more economically inactive than the actual Colombian voting-age population.
That demographic envelope is where Cepeda's vote lives. The young urban professional, the natural base for Paloma's rise and the soft anti-establishment vote that Abelardo is capturing, is structurally invisible to this methodology.
Not under-represented. Invisible. Those people are at work, at the gym, telling the surveyor they don't have time. They become the 72.5%.
- Age 55 and over: 26.3% of the sample
- Strata 1 and 2 combined: 51.4%
- Homemakers, retirees, and unemployed combined: 30.0%
- Rural: 22.4%
This sample skews older, lower-strata, and more economically inactive than the actual Colombian voting-age population.
That demographic envelope is where Cepeda's vote lives. The young urban professional, the natural base for Paloma's rise and the soft anti-establishment vote that Abelardo is capturing, is structurally invisible to this methodology.
Not under-represented. Invisible. Those people are at work, at the gym, telling the surveyor they don't have time. They become the 72.5%.
The poll itself confirms the territorial dynamic. Page 53, asked of all 3,800 respondents:
FARC dissidents and the ELN are ideologically aligned with the Petrismo project, have publicly opposed right-wing candidates throughout 2024 and 2025, and have a documented track record of pressuring local populations on electoral matters going back decades.
Law 2494 of 2025 made municipality inclusion in national polls mandatory according to specific regional and population rules. The result is that Invamer's sample now contains respondents from places like:
- Litoral del San Juan and Medio Atrato in Chocó
- Tumaco and Cumbal in Nariño
- Saravena in Arauca
- Puerto Guzmán in Putumayo
- Uribe and Vistahermosa in Meta
- Cáceres in Bajo Cauca
- San Martín de Loba in Bolívar.
Every single one of those municipalities has documented presence of armed groups with an electoral preference.
The 58.1% figure is the poll telling on itself. The respondents in those territories know what is happening, and 58.1% of them are reporting it on the record to a surveyor. The published cross-tabs cannot correct for the underlying coercion. They aggregate it into Cepeda's first-round total and call it methodology.
FARC dissidents and the ELN are ideologically aligned with the Petrismo project, have publicly opposed right-wing candidates throughout 2024 and 2025, and have a documented track record of pressuring local populations on electoral matters going back decades.
Law 2494 of 2025 made municipality inclusion in national polls mandatory according to specific regional and population rules. The result is that Invamer's sample now contains respondents from places like:
- Litoral del San Juan and Medio Atrato in Chocó
- Tumaco and Cumbal in Nariño
- Saravena in Arauca
- Puerto Guzmán in Putumayo
- Uribe and Vistahermosa in Meta
- Cáceres in Bajo Cauca
- San Martín de Loba in Bolívar.
Every single one of those municipalities has documented presence of armed groups with an electoral preference.
The 58.1% figure is the poll telling on itself. The respondents in those territories know what is happening, and 58.1% of them are reporting it on the record to a surveyor. The published cross-tabs cannot correct for the underlying coercion. They aggregate it into Cepeda's first-round total and call it methodology.
Bottom line. I'm betting real capital against this poll.
If Invamer's methodology is right and Cepeda wins every runoff scenario the way the headlines say, there's free money waiting for whoever wants to take the other side. The market is open. Put your capital where their pollster's mouth is.
polymarket.com/event/colombia…
If Invamer's methodology is right and Cepeda wins every runoff scenario the way the headlines say, there's free money waiting for whoever wants to take the other side. The market is open. Put your capital where their pollster's mouth is.
polymarket.com/event/colombia…
• • •
Missing some Tweet in this thread? You can try to
force a refresh
